Edtech’s Second Attempt: Can AI Deliver the Learning—and Profits—the First Wave Promised?

The tutor can answer another question without scheduling another class. The business still has to find someone willing to pay.

That is the opportunity—and the unresolved equation—behind India’s emerging crop of AI-powered education companies.

Arivihan’s latest funding supplies an entry into the story. The startup has raised $10 million from existing investors Accel and Prosus Ventures, alongside another $200,000 from existing GSF angel investors, taking total funding beyond $15 million. Its proposition includes interactive lessons, doubt-solving and personalised study plans without relying on live teachers.

The promise is potentially consequential: individual attention without the cost of assigning a teacher to every learner.

Indian edtech has sold the promise of personalised education before. This time, investors need evidence that the technology changes both learning and the economics of delivering it.

THE CLASS THAT LOST ITS WAY

The pandemic moved classrooms onto screens and encouraged education companies to expand rapidly. Reopening exposed a difficult assumption: emergency demand for online instruction would not automatically become a lasting preference. Schools and coaching centres returned, businesses cut staff, and some platforms closed.

Lido, Udayy and FrontRow were among the casualties. Byju’s, valued at $22 billion in 2022, entered insolvency proceedings in 2024 amid debt disputes, governance allegations and boardroom upheaval. Its problems went beyond a change in learning habits, but its reversal became the sector’s most visible warning.

The correction continued to shape funding. Indian edtech recorded 31 funding deals worth $249 million in 2025, down from 48 deals worth $572 million in 2024. Those are transactions, rather than counts of distinct funded companies.

The demand for education survived. Investor confidence in particular ways of selling it suffered.

A DIFFERENT BATCH

Recent funding announcements suggest selective backing for a different proposition.

SpeakX raised $16 million in October 2025 for AI-powered English learning. Previously a children’s hobby-learning platform called Yellow Class, it pivoted towards conversational practice. Seekho raised $28 million in September 2025 for its short-form learning business, including investment in AI-powered experiences. Berkeley-based SigIQ.ai announced $9.5 million in April 2025 for personalised tutoring, with Indian competitive-exam preparation among its applications.

Together with Arivihan’s latest round, those announcements represent $63.7 million. This is an illustrative India-facing cohort, not a comprehensive sector funding total.

Their differences matter. Speaking practice, school examinations and short-form practical learning involve different customers, prices and measures of success.

The shared hypothesis is that technology can deliver more responsive learning without requiring human teaching time to increase proportionately.

THE ECONOMICS OF ANOTHER QUESTION

A live tutor has a finite schedule. Software can potentially serve additional learners at a much lower incremental cost.

AI extends that possibility beyond recorded lessons. It can generate explanations, respond to doubts and adjust practice to a learner’s difficulties.

For students who cannot afford frequent individual tuition, that could be meaningful. Vernacular capabilities could also make products more accessible beyond English-speaking urban audiences.

Arivihan says approximately 80% of its subscribers come from tier-three cities and rural India. Its planned expansion includes more states, a wider CBSE offering and additional language capabilities.

But lower delivery costs do not guarantee better learning. Explanations need to be accurate, aligned with the curriculum and pitched at the student’s level. A confident answer that teaches the wrong method can damage the product’s central promise.

Verification, content development, computing and support remain business costs. The relevant measure is the cost of helping a student learn reliably.

THE SALE STILL NEEDS A HUMAN ANSWER

Arivihan also plans to expand digital marketing and on-ground distribution.

That detail deserves as much attention as its AI research.

Automating a tutor does not eliminate the work of persuading parents to trust it. A family must understand the offering, believe it will help and choose it over tuition, coaching, free videos or other apps.

Those acquisition costs can overwhelm an efficient delivery model.

Arivihan’s latest disclosed figures illustrate the distance still to travel. FY25 revenue increased nearly fifteenfold to ₹3.14 crore, while losses stood at ₹4.45 crore. FY26 accounts remain unavailable in the supplied reporting.

Rapid growth from a small base establishes momentum. It does not yet establish sustainable economics.

The next evidence should include how much acquiring a paying student costs, how long students remain active and whether subsequent payments cover both acquisition and service expenses.

DUOLINGO OFFERS A WORKED EXAMPLE

A global comparison shows what measurable progress can look like.

Duolingo reported approximately $33.2 million in net income in the June quarter of 2026. Its shareholder letter linked margin performance partly to AI cost efficiencies and a measured expansion of AI-powered features.

That demonstrates a connection between AI deployment and commercial economics. It does not establish that AI created Duolingo’s profitability, or that an Indian exam-preparation startup can reproduce its model.

Duolingo already has an established brand and learning habit. Younger companies must build those advantages while funding product development and distribution.

Its example makes the question more demanding: can AI improve an education product while keeping its costs controlled, rather than simply adding another expensive feature?

WHO OWNS THE ADVANTAGE?

There is another complication. AI tools are available to competitors too.

If every platform can generate explanations and conversational practice, access to the underlying technology becomes a weak basis for differentiation.

More durable advantages could include curriculum-specific accuracy, strong regional-language delivery, evidence of learning improvement and a trusted relationship with students.

Engagement metrics need interpretation as well. Time spent chatting with a tutor is useful only when it advances learning. An app that encourages dependence on answers may produce activity without understanding.

The new cohort must therefore demonstrate two outcomes together: students gain something worth paying for, and the company earns enough to keep providing it.

THE EXAM IS COMMERCIAL

AI gives education companies a credible opportunity to change how personalised help is delivered. It could widen access and reduce dependence on scarce teaching hours.

It cannot repair weak governance, undisciplined expansion or acquisition spending that never pays back.

Arivihan and its peers deserve assessment on their own results. Their funding shows investors are willing to finance another experiment.

A different chapter will be written through reliable learning, repeat payments and improving economics. The technology can answer the student’s next question. The company must still answer the investor’s: when does this become a sustainable business?